The PerformanceTests test harness treats any stderr output as a test failure. Instead, it should treat it like the normal test system, where stderr is captured and noted, but does not contribute to the pass or fail status of a test.
<rdar://problem/58767355>
IIRC Ryosuke found it very important to catch errors on this way. Perhaps this is less relevant after redoing much of the logging system?
(In reply to Brent Fulgham from comment #0) > The PerformanceTests test harness treats any stderr output as a test > failure. Instead, it should treat it like the normal test system, where > stderr is captured and noted, but does not contribute to the pass or fail > status of a test. I don't think we want to do this. Some of our internal benchmarks get broken all the time because we ignore all error messages.